Guidelines provide a means for organizing information relevant to a particular decision, such that the decisionmaker is helped by the structuring of empirical data, scientific analysis of the meaning of data, and the screening out of irrelevant information. The decisionmaker is thus left to deal with the unusual elements which might give good cause to vary a decision from the guidelines. There are three essentially different methods of guideline development. One is the model which starts from an empirical study that analyzes existing decisionmaking policy and then corrects any policies deemed undesirable. This approach allows for gradual progress that builds from the existing system. Another approach examines what decisionmakers believe underlies their decisions. Such an approach begins by examining all facts the decisionmakers believe are used in making their decisions. The first analysis would determine whether the decisions could be predicted from the subjective assessments of various factors. Variations in what decisionmakers believe to be the basis for their decisions would be noted and corrections could be made through guidelines. The third model begins by determining what ought to be the basis for decisionmaking, without regard to what is or is not believed to be occurring currently. Such 'ought' models are similar to mandatory sentencing, in that the 'correct' decision is supposedly produced by following the prescribed guidelines. This approach tends to restrict the value of discretion. Diagrams illustrating guideline schemes and a chart depicting items of the salient factor score are included.